1. Executive Summary & Core Technology
Traditional barcode scanning apps fail Indian consumers. The retail landscape features unbranded items, regional variants, loose goods, and frequent formulation changes that bypass standard global barcode databases.
COCALIT solves this through Live Shopping Mode: a multimodal computer vision and conversational voice assistant. Operating entirely via the user's smartphone camera, it visually inspects grocery store products on the shelf and verbally guides the consumer in real-time on what to buy or avoid, tailored to custom profiles (Vegan, Jain, Diabetic, specific allergens).
2. Business Model: B2B FMCG Data Licensing
While the consumer-facing app is free to use, COCALIT operates a high-value B2B data licensing model targeting FMCG conglomerates.
The Blind Spot We Solve: FMCG brands spend millions trying to understand why a consumer picks up a product from the shelf but puts it back before reaching the billing counter.
- Shelf-to-Cart Analytics: Our camera vision acts as an active data loop. We capture real-time, anonymized rejection data—identifying exactly which ingredient or dietary flag caused a consumer to return a product to the shelf.
- Unannounced Formulation Tracking: Our active video streams instantly flag when a manufacturer silently changes ingredients or packaging, providing real-time market intelligence to researchers and brands.
3. Development Stage & Traction
COCALIT is currently a pre-seed stage startup moving through the deep-tech development phase, preparing for formal incorporation in Bangalore, India.
- Model Pre-Training Phase: We are actively training our custom vision and text analysis models on a foundational, field-verified dataset of over 20,000+ Indian grocery SKUs.
- Field Testing: Video streams are being collected across varying retail lighting conditions, shelf clutter, and camera glares to ensure real-world resilience before consumer rollout.
- Fundraising: Currently initiating preliminary conversations with angel investors and pre-seed venture funds to scale backend infrastructure and engineering hires.
4. Core Architecture & Multimodal Pipeline
To process live video feeds with near-zero latency, heavy inference is offloaded from the mobile device to Google Cloud high-performance endpoints.
- Vertex AI & Compute Engine: Hosting, fine-tuning, and running our custom vision and OCR inference models at low latency.
- Google Cloud Run & GKE: Managing scalable, containerized backend microservices to handle concurrent video data streams from active shoppers.